Open Innovation Campus (OIC) Business Model
M s . K a r e n R o t h
A F R L / R I C h i e f E n g i n e e r
M r . S t e p h e n C o l e n z o
A F R L / R I O R T A
DISTRIBUTION A. Approved for public release: distribution unlimited. Case #88ABW-2019-3222.
M r . M i c h a e l W e s s i n g
G r i f f i s s I n s t i t u t e
DISTRIBUTION A. Approved for public release: distribution unlimited. Case #88ABW-2019-3222.
Agility Innovation Partnershipsare ALL required to compete in today’s exponentially
changing technology landscape!
In order to enable superior collaboration, the AFRL Information Directorate
is developing an Open Innovation Campus to provide an
enhanced environment fostering discovery and technology advancement through
fundamental scientific research in partnership with the Griffiss Institute.
Goal and Strategy
• Goal: Develop an agile and transformative ecosystem at Griffiss Technical Park,
connecting global technology leaders to collaborate and solve complex Air Force
computing challenges. Linking researchers from government, industry, and academia,
to share the best and brightest people, ideas, and facilities virtually and in person.
• Strategy: Create an innovation eco system outside of the AFRL boundaries for external
outreach and collaborations, combining the PIA partners, commercialization, open
campus, and other open innovation models together in a business model
"We will emphasize new skills and complement our current workforce with information experts, data scientists, computer programmers, and basic science researchers and engineers-to use information, not simply manage it. “ -Secretary of Defense, James Mattis - 2018 National Defense Strategy
“Evaluate service pilots similar to the U.S. Army Research Laboratory’s Open Campus, potentially expanding engagement and formally integrating them into Air Force procedures” – U.S. Air Force 2030 Science and Technology Strategy
Open Campus Concept of Operations
Step 1: Challenge Statements
• A key component to the success of the OIC is having a set of challenge problems to engage potential partners on versus assuming “If we build it, they will come”
• The challenge problems (next slide) were created in conjunction with the Core Technical Competency (CTC) leads at AFRL/RI as areas that would compliment their strategic visions but we don’t have active work in today – e.g. Seedling areas for future work
• Envision that this is updated yearly around 1 January to validate challenge problems
• Will distribute to research partners
Challenge Statements• What AF problems can take advantage of quantum information processing and how do we optimize these problems to the ever scale
quantum hardware? - Quantum Algorithm Optimization and Development on Prototype Quantum Computers with IBM, Google, QCWare, etc.
• In the World of IOT, what alternative hardware is needed to support machine learning? - Neuromorphic Computing Hardware for Size, Weight,
and Power Limited Systems - Much of current machine learning is focused on software, but especially when you consider Internet of Things devices,
we need different kinds of hardware and likely different machine learning algorithms than simply ANNs
• How can ML technology overcome adversarial attacks and data contamination?
• How can ML theory and methods to enable the rapid training and/or adaptation of predictors using a small number of labeled training
examples?
• How can we better characterize and manipulate software? - Can we model software, in an abstract space such that we can: Make statements
about its correctness / absence of classes of bugs? Analyze its runtime behavior? Quickly ascertain the function, operation, or effect of a piece of code
within a given system/context? Identify necessary changes/updates to software to meet a new specification?
• How do we represent and reason about a cyber situation / environment / mission? - To date, most of the work relative to machine learning in
cyber have focused on pattern recognition (e.g. attempts to recognize APTs), focusing only on one data type. Additionally, most recent advancements
in learning algorithms generally have been limited to real-valued data environments, absent other sources or types.
• How do we do distributed C2 in contested environments with verifiable composable C2 systems? - Research areas include control theory,
distributed systems, fault-tolerant systems, systems of systems, and formal methods
• What does quantification of multi-domain effects look like? - Research areas include adaptive sampling, probilistic modeling, and uncertainty
theory
Challenge Statements, cont.
• To what level can be make security (reliability, robustness) guarantees? - To date, most work on security guarantees has focused on formal
methods, which have proven themselves promising for small systems. Even with formally verified element(s) (e.g. a trusted processor or formally
verified application), there is still a need to interface with non-(formally) verified hardware, software, and networks. This condition gives rise to the
possibility of emergent security failures and inappropriate levels of trust/functionality being placed into the system.
• How can we derive and quantify meaningful, explainable indicators in cyberspace? - The space of metrics and measures in cyberspace is vast
and diffuse. Despite the vast literature on the subject, few of these measures have been subjected to rigorous and repeatable experimental validation,
leaving open the question of their validity or scope. Coupled with this is the need for measures and metrics whose purpose and meaning are
inherently explainable and easily incorporated into algorithms, analyses, and reports.
• How do we efficiently analyze models for operational domains? - Specifically looking at accuracy, performance, reinforcement learning, etc.;
improved training data
• How to we verify trust and deception? - Specifically Publicly Available Information (PAI) Veracity and Acoustic speech modeling
• How do we perform better conversational analytics and confirm trust in speech?
• What is the optimal waveform for different scenarios? - Constraints include user mobility, latency, interference, and power
• What is the optical scheduling for users to communicate? - Constraints include topologies, latency, and unknown number of users.
• How does a system learn in-mission and adapt to the mission environment? - In order to do decision making from limited data the research
areas include transfer learning, domain adaptation, self-supervision, interactive learning, reasoning under uncertainty, and active learning
Step 2: OIC Commitments Today• Quantum Information Science (Both 1st and 3rd Floor)
• Planned use for 3+ years
• Proposed partners: RIT, UMD, Syracuse, SUNY, IBM, PsiQuantum, IonQ, Austria, Australia, Canada, Singapore, UK, Netherlands – Team is working on commitments currently
• RIPE – RIE Rapid Test Cell
• First effort focused on adversarial autonomy
• Hosted at current Griffiss Insitute, currently ~5-6 Govt supporting and will ramp up to ~13 this summer with faculty supporting effort
• Exploring relationship with DSTL and other universities to join RIPE
• Mini-Kessel Run (KREL)
• Effort to spearhead replicating tool chain and environment of KREL here in Rome to be able to develop directly for the Air Operations Center (AOC)
• Project will be kicked started this summer via Directorate Cyber Vulnerability Assessment Team (DCVAT) team (w/ 6 ACE Interns) and 2 KREL placements
• ACE – Cybersecurity Bootcamp
• Primarily Summer Use (65-75 ppl), with additional servers/equipment and personnel year around
• Small Business Outreach
• AFRL/RI Small Business Office will host an open-door environment once a week in the OIC. This will enable outreach to small businesses looking to discuss AFRL contracting opportunities as well as ensure that partners in the OIC are aware of the resources the AFRL/RI Small Business Office can offer
PLUS all the great things the Griffiss Institute does today!
In Development Corporate Relationships
• SUNY Poly
• Developing workshops to be hosted in OIC via Systems Engineering Masters program
• SUNY Poly CAT on AI/ML
• If successful, will be co-located within the facility
• Syracuse University
• SU is looking to corporately reestablish defense relationships, including with AFRL
• MD5
• Established relationship with Dan Madden and MD5, one of the organizations that hosts Pitch Days, including the recent AI/ML Pitch Day in NYC that we were invited to attend
• AFRL/RI is pursuing future Pitch Day opportunities with Mr. Madden
• Mr. Madden is additionally looking at doing introductions to NYC universities to socialize challenge problems
• Universities
• Will continue to approach universities to establish opportunities to participate in no-cost efforts to mentor students and work on challenge problems, e.g. Senior Projects
• AFWERX – AF Innovation Hubs
• AFRL/RIB is actively pursuing opportunities with AFWERX, in the beginning stages currently
• Flyleaf program has an active project with AFWERX Vegas and are evaluating future opportunities
• University of Wisconsin Classes via AI/ML Center of Excellence
• AFOSR will coordinate having some of the courses/workshops from the Wisconsin AI/ML COE replicated here in Rome
Step 3: Outside the Fence Space Options
• Griffiss Institute Move
• Nearly 40k of available space split between three floors at Bldg 100 at the Oneida County
Airport
• GI is renting this space from the County and will expand their potential scope of work and
capabilities
• Keeps Closed/Closed and Open/Open concept where AFRL can expand capabilities to work
with partners while keeping secure work more inaccessible
• Allows expansion of ‘hard lab’ space in open domain research areas for collaboration with
foreign entities or non-citizens
• Gives rapid prototyping spaces and open internet domain access for open research that is not
easily accessible from the Laboratory
• New NYSTEC facility
• Will have an additional ~5k worth of collaboration space that they’ve made available for our use
to work on their PIA related activities
Current State – First Floor
• In progress – Two quantum labs
and two neuromorphic/electronic
labs
• Technical teams are actively
engaged and working with the GI to
ensure equipment and facility
support meets requirements
Current State – Second Floor
There is about 11,00 square feet of “event” space the will be
split up as follows:
• A reconfigurable Auditorium that can seat up to 250
people.
• The auditorium has a retractable wall that can split the room
into 2 separate rooms.
• Lobby meeting area outside of the auditorium
• 2 training/class rooms – one holds approximately 30
students and the other holds approximately 20 student
• 1 Small class room/board room – holds approximately 15
students
• 1 small teaming room for 4 people
• Kitchenette area to handle catering for events
• Small gathering/open area near escalator for small group
discussions or individual activities
Current State – Third Floor
• ~10 different work areas - Varying sizes so it’s
not “One size fits all” mentality
• Approximately 175 personnel can fit on
the third floor for use
• Can comfortably fit current summer
needs (~150) with room for
additional projects throughout
• GI will put a scheduling system in place to
allow employees to know where there is
space real time and schedule long term
areas for use
Activities Completed/In Progress
• Requirements for construction have been worked via AFRL/GI/Airport and
RFP for construction is in progress:
• Schedule:
• 75% submission June 3
• 90% submission June 17
• 100% submission July 2
• Bid docs July 16
• Bids due Aug 12
• Contract award Sept 3
• COO (cert of occupancy) March 3, 2020
8/14/2019 14
Active To Dos - Way Forward – Slide 1• Visit other Innovation Environments
• Evaluate current CRADA relationships – In progress
• Steve Colenzo working through current relationships to see what ongoing partnerships can benefit from
increased resources
• Create Communications Plan for Interaction with OIC partners
• Establish how will accomplish tasks jointly between teams in accordance with MOUs
• Create a strategic outreach plan in partnership with Griffiss Institute so we’re continuously looking for new
opportunities through the business model even if we’re “full”
• Evaluate current OIC “Buy-Ins” (e.g. Google, etc) for firm relationships against challenge problems
• Engage Security for Anti-Terrorism
• Security Team has done an initial walkthrough of the County facility to create a security plan and anti-
terrorism recommendations
Active To Dos – Way Forward – Slide 2
• International CRADAs
• Actively working templates and initial agreements for collaboration
• Corporate Communications Guide to OIC
• Will expand into more DISTRO A promotional material, including a welcome and visitors guide
• GI will host a permanent web solution on how to engage with Open Campus
• Logistics Plan for OIC
• Establish more robust equipment tracking mechanism with Logistics for OIC
• Ensure that research done on open internet is within regulation
• Engage NYSTEC further on use of their space and keep status of opening date
• Engage with AFRL Intrapreneur – Ryan Helbach
Frequently Asked Questions
• Can anyone get space?
• Yes, if you have a partnership with AFRL and agreement from AFRL
• How can we help?
• If there is interest in establishing a new or expanding an old partnership in alignment with the challenge
problems please bring it to our (Myself, Mr. Colenzo or Ms. Mills) attention sooner. This way we can ensure
that we’re prepared for longer term use and the equipment/logistics is ready
• Help promote the challenge problems to your contacts and let them know we’re looking at more out of the
box opportunities to work with them – Our goal is to have 8-10 active relationships ready to work in the 3rd
floor in a year
Questions?
Top Related